This renewal application continues to develop experimental, conceptual and computational approaches to understanding epigenetic drivers of cancer progression. A major discovery in the current period was that pancreatic cancer metastasis is driven by large-scale changes in the epigenomic landscape in the absence of metastasis driver mutations. We also developed a powerful computational approach to measuring the epigenomic landscape including deep properties such as entropy that drive pluripotency and the variable gene expression underlying tumor cell heterogeneity. The central question in the current application is the relationship between epigenetic modifiers (typically mutated in cancer) driving malignant changes in the epigenetic landscape, and their target mediator genes on the altered landscape (often unknown) and their relationship to gene expression and phenotypic heterogeneity. We will focus on acute myeloid leukemia (AML) because of the direct involvement of epigenetic modifier mutations in disease pathogenesis, the pressing need to identify their mediators, strong evidence of the clinical and prognostic impact of epigenetic heterogeneity in AML, the availability of a richly annotated set of primary patient samples via collaboration with Ravi Majeti (Stanford University), and the ease of experimental manipulation of AML cell lines and primary hematopoietic stem/progenitor cells (HSPC) for validation. Our first Aim is to generate comprehensive genome-wide maps of the epigenetic landscape in AML, encompassing high-dimensional properties such as DNA methylation entropy. We will: perform DNA methylation potential energy landscape analysis of primary AML samples with genetic mutations of epigenetic modifiers that affect DNA methylation or chromatin directly; perform similar landscape analysis of normal early hematopoietic progenitors; measure higher order chromatin structure to integrate DNA methylation, modifiers and mediators; and identify gene networks regulating gene expression and gene expression variability. Our second Aim is to perturb the epigenetic landscape to modify gene expression, expression variability and phenotype. We will: genetically engineer HSPCs to perturb epigenetic modifiers; target sites of differential entropic sensitivity on the epigenetic landscape affecting gene expression heterogeneity; validate functional roles for epigenetic mediators of the destabilized epigenetic landscape in AML; and determine the impact of epigenetic therapeutics on DNA methylation entropy in AML and HSPCs. This work will bring to bear powerful new mathematical and laboratory methods to integrate the epigenetic landscape with gene expression mean and variability, in normal and malignant hematopoietic cells, and enable us to understand gene networks, epigenetic instability and phenotypic plasticity in the context of experimental hematology and cancer progression.